Detect changes by taking continuously QQVGA/grayscale pictures and averaging pixels. Takes about 1.5ms to take and compare this small pictures.
After change is detected, camera mode switches to operate at UXGA/JPEG. Then takes a single picture, adds it to the queue and returns to detection mode. The queue is processed every 5 seconds.
Bellow is an average time to detect and take pictures at given resolution:
| Resolution | Time |
|---|---|
| UXGA | ≈1.5s |
| SVGA | ≈1s |
| VGA | ≈1s |
| QVGA | ≈.86s |
Note: Consider ±50ms
You can adjust the sensitivity by tweeking the config MAX_ACCEPTED_AVG_DIFF, 0 being the most sensitive and 255 the maximum.
Also you can change the ammount of pictures that goes into the queue by change config PICTURE_QUEUE_SIZE.
All camera pins are configured by default accordingly to this A.I. Thinker document and you can check then inside Kconfig.projbuild.